I split the sent patchset, please consider this part for current. I'll send a test for [1] in a day or so. Regarding [3], Jens, I haven't looked properly yet, how long splice can wait on a inode mutex, but it can be problematic, especially for latencies. How about go safe for-5.7, and maybe think something out for next? Pavel Begunkov (3): io_uring: fix extra put in sync_file_range() io_uring: check non-sync defer_list carefully io_uring: punt splice async because of inode mtx fs/io_uring.c | 20 ++++---------------- 1 file changed, 4 insertions(+), 16 deletions(-) -- 2.24.0
